Research On The Spatial And Temporal Characteristics And Geophysical Sources Of Common Mode Error In GPS Coordinate Time Series

GPS reference station coordinate time series provide valuable basic data for geodesy and geodynamics research for the study of global plate motion,geocenter motion,non-linear seasonal variation,postglacial rebound,etc.GPS coordinate time series can not only reflect the linear motion of reference stations,but also can reflect the non-linear motion caused by geophysical effects and other factors.The study of non-linear variation characteristic of GPS coordinate time series is of great significant to establish and maintain the dynamic reference frame of earth.GPS coordinate time series can be influenced by various factors,among which common mode error(CME)affects the accuracy and reliability of the estimation of the position and velocity of a reference station because CME easily covers up the interior movement characteristics of the reference station.In this paper,the temporal and spatial distribution characteristics of common mode error are deeply analyzed and the potential sources of common mode error are further discussed.The effects of geophysical signals,including environmental loading and thermal expansion,are analyzed quantitatively on common mode error.The main work and contributions in this paper are as follow:(1)The theories and analysis methods of GPS coordinate time series are introduced systematically,including achieving methods of GPS coordinate residual time series,the spectrum and noise analysis methods,the effecting mechanism of non-linear variation of GPS coordinate time series and calculation methods of common mode error.(2)The spatial distribution characteristics of common mode error are analyzed.The results show that the correlations between stations are strong in small-scaled regional networks,and the common mode errors in coordinate time series are approximately evenly distributed.Regional stacking filtering method and PCA/KLE method are suitable for calculating the common mode error in small-scaled regional networks.As the scale of regional networks increasing,the common mode errors in regional networks are no longer evenly distributed,and the spatial distribution of which is related to the correlation between stations.Correlation weighted stacking filtering method is suitable for the calculation of common mode error in middle-and large-scale regional networks.The spatial distribution of common mode error is not only related to the scale of regional networks,thus the common mode error calculated by correlation weighted stacking filtering method could have deviation even in small-scaled regional networks when the local effects are obvious.The spatial distribution of common mode error in horizontal and vertical component are different,thus the effecting mechanism of the sources of common mode error are different in horizontal and vertical component.(3)The relationship of the elimination of common mode error and Helmert coordinate transformation are analyzed based on the spatial distribution characteristics of common mode error.For small-scaled GPS networks with 500 km,the spatial distribution of common mode error in the networks is consistent,and could be considered as evenly distributed.In this condition,common mode error can be calculated by regional stacking filtering method,which is equal to Helmert coordinate transformation and can be instead by Helmert coordinate transformation.For the regional networks over the scale of 500 km,whether the elimination of common mode error can be instead by Helmert coordinate transformation depend on the size and consistency of correlation coefficients between stations.If the correlation between stations is strong and consistent,the elimination of common mode error can be instead by Helmert coordinate transformation.If the correlation between stations is weak or dispersive,the method of Helmert coordinate transformation is not suitable and correlation weighted stacking filtering method or PCA/KLE should be considered to calculate the common mode error.(4)The temporal distribution characteristics of common mode error are analyzed.Common mode error present obvious periodicity characteristic.The method of Lomb-Scargle periodogram is used to calculated the power spectrum of common mode error.The periodogram of GPS coordinate time series before and after eliminating common mode error are compared,and the results show that common mode error have the periodical characteristics of annual and semi-annual variation and 1.04 cpy signals.In addition,the elimination of common mode error can reduce the amplitude of colored noise,thus colored noise contained in common mode error.(5)The sources of common mode error are analyzed based on the temporal and spatial distribution characteristics of common mode error,and the effects of environmental loading and thermal expansion on common mode error are analyzed quantitatively.The results show that environmental loading may be one of the potential sources of common mode error.In vertical component,the effects are significant,and the correlation of environmental loading and common mode error is strong.After the correction of environmental loading,the reduction of RMS of common mode error is up to 1.5 mm.The contribution of environmental loading for common mode error is approximately 20%.However,the effect of thermal expansion on common mode error is not significant.Thermal expansion is not the potential source of common mode error.(6)The spectral index and optimal noise model of common mode error are analyzed.The spectral index of common mode error are closed to 1 in horizontal and vertical components,which shows that the main noise in common mode error may be:flicker noise.The optimal noise model of common mode error of CMONOC is WN+FN and WN+PL.In horizontal component,theses two kinds of noise is equal.But for vertical component,there are more stations contains WN+PL.